3) assumes

statistical independence among the local test results, the FDR procedure is (as will be illustrated in the following section) approximately valid (while being somewhat conservative) even when those results are strongly correlated, unlike the use of Eq.

So any estimation of probabilities associated with the TC must account for this lack of

statistical independence as well as for the nature of the distributions.

As can be seen in Tables 3 and 4, the assumption of statistical independence is contradicted by the data observed in relation to multiple variables.

Where the assumption of statistical independence is not contradicted, no differences between the distributions exist, namely with none or limited (non-significant) statistical differences, indicating no or limited relationships between the variables.

To further ensure

statistical independence, we randomize the hold-off time between measurements.

The following sections of this paper examine the nature of

statistical independence and test the

statistical independence of the two key variables, the participation rate and the employment rate.

Consequently, the kriged variance for a set of measured (independent) data and some arbitrary but not over-smoothed set of kriged (dependent) data violate the requirement of

statistical independence.

This condition leads us to a very useful measure for investigating the

statistical independence of the k-tuple consisting of successive terms of a GFSR sequence.

The reason for this is that the latter increases the possibility of the

statistical independence being violated and systematic error being created (McMillan, 1999; Onwuegbuzie, in press).